A peer-to-peer approach to large-scale content-based publish-subscribe

Publish-subscribe has become a prevalent paradigm for asynchronous communication in a distributed network. However, existing distributed content-based publish-subscribe systems still have limitations in scalability and adaptability. In this paper, we present a content-based publish-subscribe system over Distributed Hash Table (DHT) based P2P networks. To improve the efficiency of event publication, our approach maps subscriptions and events to rendezvous nodes in terms Of a combination of the domain schema identifier and the numbers of attributes in the subscriptions or the events. To impose little overhead on routing substrate, our approach exploits the embedded multicast tree on DHT-based P2P networks for event delivery. The experimental results demonstrate that our system scales well with increasing number of peers and large numbers of subscribers and events.